Vision-based truck blind area safety distance early warning method

• Main Authors: GENG CHANGXING, SHEN RENYUAN, LIU WANFU, GU HAIYANG
• Format: Patent
• Language: Chinese; English
• Published: 13.06.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

The invention relates to a truck blind area safety distance early warning method based on vision, which comprises a target tracking stage, a depth information matching stage and a danger judgment stage, in the target tracking stage, a depth camera based on structured light is initialized, a frame of image is read and sent into a multi-target tracking algorithm, and multi-target real-time tracking is performed on people and vehicles; outputting the pixel coordinates of the target on the RGB image in real time; in the depth information matching stage, target bounding box information of a current frame target on an RGB image is obtained and mapped into a depth image, an adaptive mask image is generated on the depth image, bitwise AND operation is performed on the depth image and the mask image, and a median is taken as distance information of the target; in the danger judgment stage, current state judgment and movement trend judgment are carried out, and corresponding danger alarm is triggered. According to the